U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications


Ocean Model Studies of Upper-Ocean Variability at 0°N, 160°W during the 1982–1983 ENSO: Local and Remotely Forced Response

D.E. Harrison

NOAA, Pacific Marine Environmental Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115

A.P. Craig

School of Oceanography, University of Washington, Seattle, WA 98195

Journal of Physical Oceanography, 23(3), 426-451 (1993)
Copyright ©1993 American Meteorological Society. Further electronic distribution is not allowed.

4. Remote forcing

The local experiments suggest that remote forcing west of 170°W plays a significant role in the equatorial response at 160°W in the SADLER hindcast experiment. In particular, local forcing and local plus forcing east of 160°W are not able to account for 1) the observed rapid deceleration of the EUC in July 1982 while the surface current is accelerating eastward more gently; 2) the penetration of the eastward surface jet during November 1982 and its rapid deceleration during December 1982; 3) the sharp warming between the surface and the thermocline in July 1982; 4) the shallowing and intensification of the thermocline during December 1982; and 5) the lower thermocline spreading during January 1983. In this section, results from several idealized westerly wind burst forcing experiments are presented to gain a better understanding of the model central Pacific response to such forcing. While other types of westerly variability occur, and are present in 1982, westerly wind bursts are a prominent aspect of west Pacific variability during ENSO periods (e.g., Luther et al. 1983; Harrison and Giese 1991). Both nearly linear and nonlinear model experiments will be considered. Idealized forcing is used because the actual wind variability west of 170°W during 1982 is quite complicated (Figs. 3, 4), and we were not able to devise a straightforward way to isolate particular periods of 1982 while maintaining the preexisting oceanic structure until the forcing period. We shall examine how much the initial state of the ocean under the forcing region and the conditions that prevail east of the forcing affect the central Pacific response.

a. Idealized westerly forced experiments

According to linear wave theory for a horizontally uniform, initially resting fluid with an exponential density profile in the vertical, the first several Kelvin wave modes all have the same sign of zonal velocity amplitude above the thermocline. Furthermore, the response to westerly wind events from these modes is a pulse of eastward acceleration above the thermocline and a deepening of the thermocline. Conversely, easterly wind events cause a zonal flow deceleration and shallowing of the thermocline from the lowest modes. Thus, far enough from the forcing region that the different vertical modes have separated due to their different eastward phase speeds, one expects to see a series of zonal surges and slowings as a function of depth in response to a single wind event that includes a wind extremum followed by a return to prevailing conditions. A wind change in which the wind simply decreases or increases will lead to the first part of the behavior induced by a pulselike event.

Giese and Harrison (1990) show model results for a 0.02 N m wind burst on an exponentially stratified, resting ocean (their experiment B24, which we shall denote WBex0.02, for westerly burst/exponential stratification/0.02/stress). The idealized westerly wind event is Gaussian in time with half-amplitude points ten days apart, Gaussian in latitude about the equator with half-amplitude points 3 degrees apart, and is uniform between 150°E and 170°E. The wind field is shown in Fig. 13. Giese and Harrison (1990) found that the far-field model results from this case agree well with linear theory. Kelvin modes propagate away from the forcing region at speeds consistent with linear theory, and the amplitudes of the wave perturbation as a function of depth agree with the theorized structure of the various baroclinic modes. They also noted that while modes 3 through 6 are strongly forced for zonal velocity, their sea level signal is small. We are interested in the equatorial central Pacific response.

Figure 13. Zonal wind stress along the equator for the WBex0.02 experiment (0.02 N m wind burst). Meridional wind stress is identically zero for the same period. The vertical axis is months.

The WBex0.02 zonal velocity and temperature time series at 160°W and 180°, 30°, and 10°, respectively, east of the forcing, are shown in Fig. 14. Recall from Fig. 13 that the forcing event begins early in month 1 and ends early in month 2. At 180° the zero crossing of the total zonal velocity perturbation rises to thermocline depth by the middle of month 2. It is not useful to try to identify particular modes, as most of the Kelvin modes have not separated. Linear theory says that the third and fourth modes receive the strongest zonal velocity forcing, followed closely by the fifth and the second modes; the first mode receives about one-third of the forcing that the third mode receives (Giese and Harrison 1990). The relationship between the zonal velocity perturbation seen in Fig. 14 and the thermal structure is simple; positive zonal velocity perturbations are associated with downwelling and local warming while negative zonal velocity perturbations are associated with upwelling and local cooling.

Figure 14. Zonal velocity and temperature at (0°, 160°W) and (0°, 180°) for WBex0.02 (0.02 N m burst over a resting ocean with exponential density profile in depth). The contour intervals are 2 cm s and 2°C, respectively.

As predicted, vertical modes higher than the second play an important role and lead to westward flow at depths as shallow as 40 m in the first month following the forcing event. Even from this linear perspective, forcing like this can lead to near-surface eastward acceleration simultaneous with thermocline westward acceleration, as was found in SADLER in July 1982; however, the vertical structure of the thermal changes shows no significant intensification of stratification during the shallowing phase as was found in SADLER during December 1982-January 1983. At 160°W the lowest modes are slightly better separated than at the date line, but the overall pattern of response is similar, just stretched out some in time; westward subsurface flow begins in the thermocline about the middle of month 3. We next consider the effects of more realistic forcing amplitude, vertical density stratification and background circulation.

In the next experiment (WBwp0.2), a 0.2 N m version of the same westerly wind event as used in WBex0.02 is introduced over a resting ocean in which the stratification is typical of 160°E. In the third experiment (WBcl0.2), a similar wind event is superimposed on a climatological annual mean wind stress field that has already been used to spin up the ocean to its mean circulation. These experiments are described in detail and correspond, respectively, to B32 and B34 in Giese and Harrison (1990).

The eastward phase speeds of the low vertical modes in WBwp0.2 are not far from linear predictions, based on far-field arrival times (Giese and Harrison 1990), and are substantially faster than for the WBex0.02 case. The zero crossings for zonal velocity are substantially deeper, and the zonal velocity forcing of the low vertical modes is quite different than in WBex0.02. These differences result entirely because of the stratification change. Now the strongest forcing is of the first mode, followed by the second, fourth, and then third mode. The third mode receives about half the forcing of the first mode. This simple calculation serves as a reminder of the well-known importance of stratification on the vertical structure and response of forced Kelvin motions (e.g., Busalacchi and Cane 1988; Cane and Sarachik 1976).

Figure 15 presents time series of temperature and zonal velocity at 160°W and 180°, respectively, for WBwp0.2. Note that the contour interval for zonal velocity has been changed from 2 cm s to 20 cm s, to reflect the tenfold increase in wind stress forcing. At 160°W the largest velocity change is westerly flow in the thermocline between the middle of months 3 and 4. The velocities are less than ten times those seen in WBex0.02, but the pattern is similar in form after allowing for the deeper zero crossings and faster phase speeds of WBwp0.2; the forcing projection onto the different vertical modes accounts for most of this difference. This example illustrates that stratification plays a significant role in the magnitude of the forced response as well as its vertical structure. The temperature field is generally depressed initially and shoals later, but the character of the changes depends significantly on the depth of interest; there is little near-surface displacement. At depth, rapid warming occurs, followed by significant cooling, and then warming again as more normal temperatures return. During the shoaling phase a modest increase occurs in thermocline stratification.

Figure 15. Zonal velocity and temperature at (0°, 160°W) and (0°, 180°) for WBwp0.2 (0.2 N m burst over a resting ocean with density profile typical of 160°E). The contour intervals are 20 cm s and 2°C, respectively.

At 180° there is substantial positive zonal velocity perturbation, initially between the surface and about 200 m; subsequently the zonal flow remains eastward down to about 100 m into month 3, and is westward below about 100 m until the end of month 2. The largest zonal velocity change (100 cm s) occurs early in month 2 at about 40-m depth; the maximum thermocline zonal flow change (40 cm s) occurs a little past the middle of month 2. The temperature perturbations show rapid deepening early in month 2 followed by shoaling and an increase in stratification between 28°C and the deepest temperature plotted (16°C) until the end of month 2. In month 3 the midthermocline temperature (say 22°C) flattens out while warmer isotherms shallow and cooler isotherms deepen.

At the date line a significant departure from linear Kelvin wave dynamics is near the surface during month 2, when there is a near-surface eastward jet with maximum flow of about 1 m s at 40 m. A surface jet forms directly under the wind patch and its strong zonal currents carry the jet eastward to 180° and strong associated downwelling forces the jet deeper into the water column. In this case, the jet lasts a few weeks. Giese and Harrison (1990) note other nonlinear aspects to the response to 0.2 N m forcing.

This experiment shows that remote forcing with a stronger wind event and over a more realistic initial density profile can produce thermocline evolution much more similar to that seen in SADLER than was found in the linear westerly event forcing experiment. The distance from the forcing region has a considerable effect on the character of the response, both in zonal flow and thermal evolution; the closer to the forcing, the more aspects of the thermal structure changes resemble those seen in SADLER in July 1982.

Results for WBcl0.2, in which the westerly event is superimposed on an ocean spun up under climatological mean wind stress, are shown at 160°W and 180° in Fig. 16. The response at 160°W involves acceleration of the undercurrent from about 100 cm s to over 120 cm s, followed by a deceleration to about 60 cm s between the middle of month 2 and early in month 3. Subsequent to month 3, the instability waves, which are a part of the climatological flow in this experiment, dominate the zonal flow changes. There is little near-surface zonal flow change, never more than about 20 cm s. The thermocline deepens and the near surface warms during month 2; then the thermocline shoals and the near surface returns to its initial temperature during month 3. As in Harrison and Giese (1988), the remotely forced response alters the phase of the instability waves, so that subtracting out the climatological circulation does not greatly simplify observing the remotely forced response; we know from the previous studies that the period of maximum interest is months 2 and 3, during which clear variability is evident.

Figure 16. Zonal velocity and temperature at (0°, 160°W) and (0°, 180°) for WBcl0.2 (0.2 N m burst over an ocean forced to statistical equilibrium to climatological annual mean wind stress, with a simple heat flux parameterization). The contour intervals are 2 cm s and 2°C, respectively.

At 180° the instability process is weaker, so that the remote response is more easily seen than at 160°W. Early in month 2, the zonal velocity increases about 20 cm s from the surface to about 200 m and the thermocline deepens. During the second half of month 2 a rapid zonal velocity deceleration occurs initially in the thermocline (from almost 100 cm s to 40 cm s), as well as a shallowing of isotherms and increase of stratification. During this phase the near-surface zonal velocity decrease is much smaller, typically 20 cm s. About mid-month 2 there is rapid near-surface warming. At the end of month 2 and through month 3, the zonal velocity increases eastward and conditions begin returning to normal.

The depth of the thermocline and thermocline stratification under the forcing in WBwp2.0 and WBcl2.0 are similar, yet some aspects of the remote response are quite different. The undercurrent deceleration in WBcl2.0 is much greater; the zonal flow change is larger and occurs more abruptly. The strong surface zonal flow response in WBwp2.0 is not present in WBcl0.2, yet WBcl0.2 has much larger near-surface temperature change than WBwp0.2. Another difference between the two experiments is the lack of a surface-trapped jet at 180° in WBcl2.0. Apparently the surface jet under the forcing is not advected to 180° because there is sufficiently strong westward surface flow in the climatological flow to offset the zonal advective tendency of the equatorial jet. It appears that interaction between the forced response and the background circulation alters the response considerably from that found in WBwp0.2, even though the background stratification is similar. We have now been able to reproduce each of the characteristic features of the remotely forced response in the SADLER experiment, as will be discussed more thoroughly below.

b. Forcing west of 160°W in the SADLER experiment

Now we attempt to explain the elements of the remotely forced response at 160°W in the SADLER experiment due to westward forcing. Recall the zonal velocity and thermal response differences between the LOCAL experiment and the SADLER experiment. The zonal velocity differences between SADLER and LOCAL show up as local difference extrema in Fig. 6; the major ones are late July through August 1982 at 120 m, mid-November through mid-December 1982 at about 100 m, and late January through early February 1983 between the surface and about 80 m. The EUC decelerates faster and sooner and the surface eastward jet penetrates deeper and decelerates faster in SADLER. Figure 7 shows the thermal differences in an equally clear fashion; warming begins earlier in SADLER, and SADLER remains warmer than LOCAL from the surface down into the thermocline until early December 1982 at 160 m and late January 1983 at the surface; the maximum temperature difference is about 6°C in November 1982 at 120 m. Thereafter, LOCAL is cooler than SADLER into mid-1983; differences between 2°C and 4°C are common in the upper thermocline. LOCAL does not have the rapid thermocline shoaling and intensification of SADLER in November-December 1982. The 30° water at the surface in October in the SADLER experiment is not present in the LOCAL experiment.

In order to begin to sort out the remote response, look again at the wind stress changes. The largest remote zonal wind changes occur in the June-September 1982 and November-December 1982 periods (Figs. 3 and 4). The mid-1982 remote zonal wind stress changes involve an eastward-propagating westerly anomaly that begins near 150°E in April and appears farther and farther east until about June 1982 when there is a westerly event at 170°E and 180°. Between June and July the largest monthly westerly stress changes occur and are at 170°E and 180°. The simplest approach to representing this variability would be as a westerly event or as a one-sided westerly change in June-July near the date line. Late in 1982 there is a substantial westerly event in November-December at both 170°E and 180°; a more modest event occurs at 160°E and 150°E. The zonal stress continues to strengthen toward the east through December, and becomes weakly easterly by January 1983. Thus, from the idealized perspective, there is a substantial westerly event immediately followed by a one-sided (in time) return to easterly stress. The longitudinal extent of the late 1982 wind changes is the greatest of any during this event, so the integrated forcing is the greatest.

From our idealized remote response experiments it is clear that westerly wind changes west of 170°W can account qualitatively for the differences in zonal velocity and temperature behavior in July-August 1982. There is nothing inconsistent with westerly remotely forced Kelvin processes and the deceleration of the EUC as it occurred in SADLER; reasonably near to the forcing region strongly forced higher vertical modes can provide just this type of behavior (see WBwp0.2 and WBcl0.2). Further, the extra thermocline zonal velocity acceleration in November 1982 and the thermocline uplifting and intensification during December 1982-January 1983 are also possible from westerly remotely forced Kelvin response (see WBcl0.2 particularly). The wind stress variability west of 170°W in SADLER is sufficiently complex in space and time that it is not straightforward to identify particular wind events with details of the response at 160°W in SADLER, particularly given that the presence of higher vertical modal forcing means that it can take two to three months for the response to propagate even 20° or 30° of longitude east of the forcing region. Thus, at 160°W the ocean is responding not only to forcing from, say, 160° or 170°E from about ten days to three months previous but also to forcing from 180° from about a few days to two months previous and also to its local forcing (especially for the September 1982-February 1983 period).

Two aspects of the wind field present themselves as likely candidates for forcing the mid-July to August 1982 behavior: the near-date line westerly event in July 1982 and the eastward-propagating westerly event that began near 150°E in April 1982. We have not examined the consequences of eastward propagation of a westerly event in our idealized studies; Harrison and Schopf (1984) considered one aspect, and McCreary and Lukas (1986) considered the possibility of resonant forcing from such behavior. It seems plausible that both aspects of the wind field contribute to the observed mid-July-August response, but it is worthy of note that resonant forcing is not required in order to explain the observed zonal velocity vertical structure.

Westerly remote forcing can also help account for the vertical penetration and amplitude of the surface easterly jet in November 1982 and its rapid deceleration and concomitant thermocline uplift and intensification in December 1982-January 1983. The picture is necessarily less clear than during July-August 1982 because significant local wind stress changes are playing a substantial role in the local response. Because the thermocline is deep and there is little stratification above it during this part of 1982, the WBcl2.0 experiment is most relevant. The thermocline eastward acceleration in November 1982 followed by deceleration and thermocline uplift and intensification are characteristic of the early response at 180° in WBcl2.0. Continued rising of warmer isotherms while cooler isotherms deepen, as is found in January 1982, is found during month 3 in WBcl2.0 at 180°W.


Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage